Memo
38 Chauncy Street
Suite 200
Boston, MA 02111
T 857-453-5450
F 857-453-5451
www.transystems.com
To:
Steve Dudley, SCRCOG
Jim Rode, SCRCOG
From:
Jim Wensley
Date:
June 27, 2012
Subject:
Regional Transit Data Acquisition
TranSystems has completed the task to assemble the transit data needed for future enhancements to
the transit component of the regional travel demand forecasting model. Transit ridership and service
data has been assembled for all public transit providers operating in the SCRCOG region. Base year
(2010) level of service data has been assembled so that it can be readily used as an input to the
model. Transit ridership data has also been assembled and summarized so that it can be used to help
calibrate and validate the mode share and transit assignment portions of the model.
The project included several tasks:
Task 1: Determine Data Needs - TranSystems identified the data needed regarding bus services
provided in the region by CTTransit New Haven, as well as CTTransit Wallingford, CTTransit
Meriden, Milford Transit and the multiple-operator Coastal Link service. Data on routes entering
the region and operated by Middletown Transit, Estuary Transit and the Hartford express carriers
was also desired, as was data on rail services. The needs identified included level of service data
(peak and off-peak headways) and current fare information plus ridership data, including route-level
daily ridership for bus services and station level daily ridership for rail services. Bus ridership at the
TAZ level was also needed, as well as route-to-route bus transfer data. The data to be assembled,
sources of the data, and the data output format was reviewed with SCRCOG staff before the data
was assembled.
Task 2: Data Assembly - The data was assembled by contacting the individual bus and rail
operators, as well as Connecticut DOT. Schedule and fare data was generally obtained from agency
web pages since service levels have not changed on most services since 2010. Route-level ridership
and stop-level ridership was collected from CTTransit New Haven and CTTransit Wallingford. Only
Route-level ridership has been obtained from the Coastal Link but stop-level data is still expected.
Only Route-level ridership was available from CTTransit Meriden and Milford Transit. Farebox data
from which bus route transfer data could be estimated was obtained from CTTransit New Haven for
a full week in April 2010. Transfers counts were not available from other bus operators. GIS files of
bus stop locations were obtained from CTTransit and a GIS file of TAZs was obtained from
SCRCOG. Rail data was gathered from ConnDOT for Shore Line East, Metro-North New Haven
Line, and Amtrak. Wherever possible, data was assembled for year 2010.
Task 3: TAZ Estimates – Using the GIS files of bus stops and TAZs TranSystems was able to
identify in which TAZ each bus stop is located. For the majority of stops, the stops had to be
assigned to two or more TAZs, since many stops lay on the boundary between TAZs, and the
ridership had to be split among them. The process for doing this is detailed in the attached
documentation. The ridership obtained from those operators who could provide stop-level data was
adjusted to represent a full weekday of ridership. The stop-level ridership was then summed by TAZ
to obtain daily transit ridership by TAZ.
Task 4: Transfer Data - A route-to-route matrix of transfers for CTTransit’s New Haven
Division was created through manipulation of detailed farebox data. One week of cash boardings
data and one week of pass boardings data was used to develop a daily transfer matrix that can be
used for future model validation. The process followed is detailed in the attached documentation.
Task 5: Documentation - The data assembled and prepared for use in the regional model is
provided in three spreadsheet files, consistent with the format reviewed with SCRCOG staff in Task
1. The data assembled, the procedures used to prepare the data in its final format, and a description
of the final output format in included in the documentation that follows.
Regional Transit Data Acquisition
South Central Regional Council of Governments
Documentation of Transit Data
Route-Level Data
Data Sources:
•
•
•
•
•
•
•
•
•
•
Online schedules for CTTransit Hartford, New Haven, Wallingford and Meriden (headways and
fares).
Online schedule for Milford Transit, Estuary Transit, Middletown Area Transit, and Greater
Bridgeport Transit (headways and fares)
Online schedules for Shore Line East, Metro-North New Haven Line and Amtrak (headways and
fares).
CTTransit “Route Summary Productivity 2010”. Includes estimates of 2010 weekday passengers by
route for CTTransit New Haven factored from a sample of trips.
Tabulations of farebox data from October 2010 provided by CTTransit Meriden (CTTransit
Meriden ridership).
Nelson Nygaard “Wallingford Bus Ridership and Routing Study” Wallingford ride check data
summary spreadsheet. (CTTransit Wallingford ridership from September 2011)
AECOM “Coastal Corridor Bus Study, Recommended Service Plan” January 2012 (2011 Costal Link
ridership).
E-mail from Milford Transit with 2009/2010 average weekday ridership by route for Milford Transit.
Shore Line East, Metro-North and Amtrak station boardings through e-mail from CTDOT
o Shore Line East information from 2010 Daily Rider monthly reports
o Metro-North information from 2010 TOTAL WEEKDAY BOARDINGS BY STATION
o Amtrak information form Amtrak Fact Sheet, Fiscal Year 2010 State of Connecticut
Ridership data for the following routes was not available:
o Meriden-Hartford Express (Route 19) operated by Peter Pan
o New Haven-Hartford Express (Route 20) operated by DATTCO
o Middletown Area Transit M-Link
o Estuary Transit Shoreline Shuttle
Processing:
•
•
Current (2012) headways were used for all services. CTTransit indicated that the only bus route
that has changed service levels since 2010 is the S Route, which was operated at a lower frequency
of service in 2010 before being taken over by CTTransit in 2011.
Ridership is from 2010 for CTTransit New Haven and Meriden, Milford Transit and the rail services.
Ridership from 2011 was used for CTTransit Wallingford and the Coastal Link.
Output Table:
•
•
Data is contained in “SCRCOG Model Data.xlsx”.
Average weekday ridership, peak and off-peak (midday) headways, and base fare for each bus route
operating in the SCRCOG region are listed in the table. Data is for 2010 where possible; however
some data are more recent.
Regional Transit Data Acquisition
South Central Regional Council of Governments
Documentation of Transit Data
•
For rail services, average weekday station boardings are given along with line headways and fares to
New Haven (for Shore Line East and Amtrak from Meriden and Wallingford) and fares to New York
on Metro-North from New Haven.
Regional Transit Data Acquisition
South Central Regional Council of Governments
Documentation of Transit Data
Transit Ridership by TAZ
Data Sources:
•
•
•
•
•
•
CTTransit “NH Weekday Route Profile 2010” text file. Consists of a text file version of
CTTransit’s Route Profile Report. Report includes trips checked from 1/1/10 through 12/31/10.
Most scheduled trips were checked once during the year. The report shows total ons and offs
recorded by stop by time period on each route and direction for a typical day. Stops were
identified only by a text description; there was no numeric ID or alphanumeric stop code.
Nelson Nygaard “Wallingford Bus Ridership and Routing Study” Wallingford ride check data
summary spreadsheet. Contained typical weekday ridership by stop for CTTransit Wallingford
collected in September 2011.
AECOM Coastal Link data files
Stop level data was not available for CTTransit Meriden or Milford Transit
CTTransit “New Haven Bus Stops March 18, 2012” GIS shape file. Contained geocoded locations
of all bus stops in the CTTransit New Haven Division (including CTTransit Wallingford and Meriden
stops) as of March 18, 2012. CTTransit does not maintain historical GIS records of bus stops so
there was no way to get a 2010 version. The file contained text descriptions and a numeric ID for
each stop.
SCRCOG TAZ file. Geographic file of all TAZs in the SCRCOG region in TransCAD format.
Processing:
•
The route profile text file report was imported into a spreadsheet and parsed to extract the
number of scheduled trips and the number of observed trips by route, direction and time period. In
most cases 100% of scheduled trips were sampled. With the exception of the M Route, all sample
sizes were at least 67% (M Route samples ranged from 17% to 88%). The number of complete
samples and smallest sample sizes by period are shown below:
Period
Time
AM Peak
Mid Peak
5:30 – 8:30 AM
8:30 AM – 2:30
PM
2:30 – 5:30 PM
5:30 PM – 5: 30
AM
PM Peak
Off Peak
•
•
Route/Dirs.
Operated
47
43
Route/Dirs
with 100%
Sampled
35
35
Smallest
Sample
(excl. M)
67%
83%
46
43
32
36
78%
83%
The text file was imported into another spreadsheet and parsed to extract the number of recorded
ons and offs by stop for each route, direction, and time period.
In cases (i.e. route/direction/period) where fewer than 100% of trips were sampled, the observed
ons and offs were factored up to represent a full typical day using the ratio of trips scheduled to
trips observed.
Regional Transit Data Acquisition
South Central Regional Council of Governments
Documentation of Transit Data
•
•
•
•
•
The text descriptions of stops in the profile data and in the GIS bus stop file were different length
fields so the descriptions in the GIS file had to be shortened and have some spaces removed in
order to match the route profile descriptions. The edited GIS file stop descriptions were then used
to lookup and attach the numeric stop ID to each stop in the route profile data.
Because the GIS file was more recent than the route profile, there were 59 stops in the route
profiles that were not in the GIS file; however only 18 of these had non-zero ridership data. These
were assigned an unused ID number and were addressed separately later.
CTTransit Wallingford stop level ridership and Coastal Link stop level ridership were appended to
the CTTransit New Haven data.
The GIS bus stop file was converted to TransCAD and overlayed with the TAZs. (Several versions
of the bus stop file had to be obtained from CTTransit before receiving once that could be
successfully overlayed.)
Because it was noted than many bus stops lie along the streets that form the boundaries between
TAZs, TransCAD was used to determine the number of TAZs that lie at least partially within 200
feet of each bus stop. This distance was chosen after examining the number of stops serving one,
two, or three or more TAZs for several longer and shorter distances. This distance was found to
identify stops lying along the dividing street between two TAZs even if the stop wasn’t perfectly
coded along the line. It also appears to identify stops serving three or more TAZs only if the stop is
really at the junction of three or more zones. The table below shows the number of stops identified
to serve each number of TAZs.
Number of
TAZs
0
1
2
3
4
•
•
•
•
•
•
Number of
Stops
300
1,457
1,975
492
124
Stops serving zero TAZs were confirmed to be stops lying outside the SCRCOG region on the J and
F routes.
TransCAD was used to automatically tag each stop with the ID number(s) of up to two TAZs (that
were within 200 feet). The stops serving three or more TAZs had to be checked manually to add
the third and fourth TAZ ID.
The 19 stops not in the GIS file were manually assigned to one or more TAZs based on the stop
description and their order in the stoplist for the route.
Stops on the Milford portion of the Coastal Link were manually assigned to one or more TAZs
based on the stop description.
The number and list of TAZs served by each bus stop was joined to the bus stop ridership from the
route profiles using the bus stop ID number.
The ridership (ons and offs by time period) for each stop was then split evenly between the TAZs
served by that stop. While more sophisticated methods of allocation using population data and size
Regional Transit Data Acquisition
South Central Regional Council of Governments
Documentation of Transit Data
•
of the catchment area within each zone could be envisioned, these would require substantial
geographic analysis involving over 2,500 stops. Splitting the ridership evenly seems to be a
reasonable approximation, especially given that most cases with multiple TAZs were simply stops on
a straight line dividing two zones.
Matrices were then produced of ridership by TAZ and route, and ridership by TAZ and time period.
Output Tables:
•
•
Several matrices of ridership by TAZ and route were produced and are included in “SCRCOG
Transit Ridership by TAZ.xlsx”. Rows represent TAZs listed by their ID numbers. Columns
represent the CTTransit New Haven and Wallingford routes. Data represents a typical weekday in
2010. Each matrix is on a separate tab:
o Daily Route Riders – Daily ridership (total ons; offs can be assumed to be similar)
o AM Route Ons – AM peak (5:30-8:30) boardings
o AM Route Offs – AM peak (5:30-8:30) alightings
A fourth tab (Period Riders) shows ridership (both ons and offs) by TAZ and time period. TAZ
data in this table is summed over all routes.
Regional Transit Data Acquisition
South Central Regional Council of Governments
Documentation of Transit Data
Transfer Data
Data Sources:
•
•
•
CTTransit “New Haven Transaction Details – Transfer” data from fareboxes for April 9-13, 2010.
This text file contains a record for each cash transfer issued and each transfer received. Records for
transfers received contain the route codes for both the issuing and receiving route.
CTTransit “New Haven Transaction Details – Period Pass” data from fareboxes for April 9-13,
2010. Spreadsheet contains a record for each boarding using a time-based pass (e.g. a monthly
pass).
Similar data was not available for CTransit Meriden or Wallingford, or for Milford Transit.
Processing:
•
•
•
•
•
•
Only the “Transfer Received” records were used for cash-based transfers. The two routes codes
(issuing and receiving) on these records were checked for validity. Incorrect codes were re-coded
to “999”. Valid codes included CTTransit New Haven (issuing and receiving) and CTTransit
Meriden (issuing only) routes. Transfers from CTTransit Wallingford did not appear in the data. All
records were tabulated into a route-to-route matrix for cash transfers for one week. The data for
transfers issued in Meriden were duplicated in the opposite direction to account for transfers issued
on CTTransit New Haven and likely received on CTTransit Meriden.
Period pass data included records for every boarding using a pass. This data was sorted by pass ID,
date and time. Two consecutive boardings were considered a “pass transfer” if they met the
following four criteria:
1. They were made using the same pass ID
2. They were made on the same date
3. They were made within 90 minutes of each other (the valid window for “cash transfers”)
4. They were on different routes
Adjustments were made in cases where the direction for the originating trip was “outbound” or the
receiving trip was “inbound”. These are believed to often be cases where the driver switched the
route entered in the farebox to the opposite half of the same route (e.g. switching from B Whalley
to B Congress Avenue) either before the initial boarding or after the transfer boarding. In these
cases the route was swapped to the other “half-route”. (The methodology was copied from the
methodology used to create a similar transfer matrix for the 2008 Implementation of the Regional
Transit Study done by TranSystems.) All records were then tabulated into a route-to-route matrix
for “pass transfers” for one week.
The Cash and pass transfer matrices were added together.
Cash transfers to and from the Meriden routes were factored up using the systemwide ratio of total
transfers to cash transfers (1.39) since the pass transfer data did not include Meriden routes.
The figures in the combined weekly matrix were divided by five to yield an average daily transfer
estimate for the week analyzed.
Regional Transit Data Acquisition
South Central Regional Council of Governments
Documentation of Transit Data
Output Table:
•
A single weekday transfer matrix is provided in “SCRCOG Region Transfers.xlsx” representing an
average weekday in a typical week in April 2010. The rows represent the originating routes and the
columns represent the receiving routes.